June 29, 2026
Repo gossip for your AI
Lore – give your coding agent the decisions your team made
This tool wants to stop AI from forgetting your team’s old decisions and commenters are already split
TLDR: Lore is a new tool that tries to stop AI coding assistants from repeating old arguments by showing them the team’s saved decisions in the project files. Commenters immediately debated whether that’s a breakthrough or just a dressed-up version of writing better notes and pointing the bot at them.
A new project called Lore is pitching itself as the office historian your coding bot desperately needs. The promise is simple: keep your team’s past choices, plans, and rules in plain text files inside the project, then let AI coding helpers read them so they stop barging in and “fixing” things your team already decided. The creator jumped into the comments with a very relatable complaint: agents kept resurrecting ideas the team had already killed, like a zombie coworker with commit access.
That instantly set off the crowd. One camp basically said, “Wait, isn’t this just a fancier version of writing your decisions down and pointing the AI at them?” The most blunt comment reduced the whole thing to a three-step life hack: write decision notes, save them in git, mention them in your agent instructions. Ouch. Another commenter asked the obvious question: how is this really different from a simple rules file like CLAUDE.md? That’s the core drama here — game-changing system of record or just extra paperwork with better branding?
Then came the side quests. One builder popped up to say they’re making a rival-ish tool with a totally different philosophy, because of course no Hacker News thread is complete without a “funny you mention that, I built my own” entrance. And the comic relief award goes to the drive-by comment about “bad timing either way epic games release,” which landed like a chaotic meme from another dimension. In short: Lore’s big idea is orderly memory for AI, but the comments turned it into a debate over whether this is the future of teamwork or just ADRs in a trench coat.
Key Points
- •Lore stores a team’s requirements, decisions, designs, roadmaps, and prompts as typed Markdown in a repository and serves them read-only to coding agents over MCP.
- •The product is built on the open-source RAC engine, with the package, CLI, and MCP server distributed under the `rac` name.
- •Lore is positioned as a deterministic system of record, contrasted with similarity-ranked retrieval approaches such as RAG or agent memory.
- •The workflow includes CI enforcement through `rac validate` and `rac gate`, which can reject malformed artifacts, ambiguous links, and references to superseded decisions.
- •The article states the engine makes no LLM calls and no network calls by default, with only an off-by-default consent-gated usage ping mentioned as possible egress.